While the majority of children diagnosed with acute lymphoblastic leukemia (ALL) now achieve long term survival on contemporary treatment regimens, more than 20% relapse or die of resistant disease. ALL thus remains the leading cause of cancer death in children. This is particularly true for children diagnosed with 'high-risk" ALL, accounting for 30% of all cases;outcomes also remain poor in adolescents and adults with ALL. From comprehensive molecular analyses of >900 high-risk ALL cases, we have derived and validated gene expression signatures ("classifiers") predictive of relapse-free survival (RFS) in ALL and we have discovered novel underlying genetic abnormalities (IKAR0S/IKZF1, JAK, and CRLF2 mutations or genomic rearrangements) that serve as new diagnostic and therapeutic targets for this disease. We now propose to translate these RNA and DNA-based molecular signatures into robust clinical diagnostic tools;complete the validation of the signatures measured in this fashion in a new cohort >500 high-risk ALL patients and develop new algorithms for improved hsk classification, outcome prediction, and therapeutic targeting;and, prospectively test these signatures in the next generation of COG trials. Our aims are to: 1) refine our microarray-based gene expression classifiers to a final set of predictive genes and translate the quantitative measurement of this multi-analyte signature to a robust clinical diagnostic platform (ABI TaqMan(r) automated, quantitative RT-PCR cards);2) validate the predictive power of this classifier in an independent cohort of 500 high-risk ALL cases relative to the newly defined prognostic molecular abnormalities (involving IKAROS, JAK, CRLF2), and known risk factors (such as minimal residual disease) to develop a new molecular algorithm for risk classification and therapeutic targeting;3) prospectively test the predictive power of the new molecular signatures and risk algorithms in the next generation of COG clinical trials opening in 2011 which will accrue >3000 children with high-risk ALL;and 4) determine the predictive utility of these classifiers, molecular signatures, and risk algorithms in adolescent and young adult ALL patient cohorts as well as older adults with ALL accrued to clinical trials conducted by the adult NCI Cooperative Groups.